Merge pull request #1472 from BRCAChallenge/repair_vrids

Repair vrids

pipeline/Makefile

@@ -47,7 +47,8 @@ start-local-uta: ## starting local uta docker container
 	[ `docker ps -f name=$(UTA_CONTAINER) | wc -l` -gt 1 ] || docker run -dit --name $(UTA_CONTAINER) -p $(UTA_PORT):5432 $(UTA_DOCKER_IMAGE)
 
 start-seqrepo-rest-service:
-	[ `docker ps -f name="seqrepo-rest-service" | wc -l` -gt 1 ] || docker run --name seqrepo-rest-service --detach --rm -p 5000:5000 -v $(SEQ_REPO_DIR):/usr/local/share/seqrepo/ biocommons/seqrepo-rest-service /usr/local/share/seqrepo
+	[ `docker ps -f name="seqrepo-rest-service" | wc -l` -gt 1 ] || \
+	utilities/lauch_seqrepo.sh
 
 .ONESHELL:
 setup-files: ## setup various directories to run pipeline

pipeline/utilities/launch_seqrepo.sh

@@ -0,0 +1,11 @@
+#!/usr/bin/env bash
+
+SEQ_REPO_DIR="${1:-/usr/local/share/seqrepo/latest}"
+
+#
+# Launch the seqrepo rest API docker container
+docker run --name seqrepo-rest-service \
+       --detach --rm -p 5000:5000 \
+       -v ${SEQ_REPO_DIR}:/mnt/seqrepo \
+       biocommons/seqrepo-rest-service \
+       /mnt/seqrepo

pipeline/vr/appendVRIds.py

@@ -6,6 +6,8 @@
 import csv
 import socket
 import urllib3
+import subprocess
+import time
 
 from ga4gh.core import sha512t24u, ga4gh_digest, ga4gh_identify, ga4gh_serialize
 from ga4gh.vrs import __version__, models, normalize
@@ -15,7 +17,7 @@
 csv.field_size_limit(10000000)
 
 SEQREPO_REST_SERVICE_URL = "http://localhost:5000/seqrepo"
-#SEQREPO_REST_SERVICE_URL = "https://services.genomicmedlab.org/seqrepo"
+
 DP = SeqRepoRESTDataProxy(base_url=SEQREPO_REST_SERVICE_URL)
 TLR = Translator(data_proxy=DP,
                  translate_sequence_identifiers=True,
@@ -88,7 +90,10 @@ def get_vrs_id(hgvs, max_repeats=5):
             continue
         else:
             allele_dict = allele.as_dict()
-            return(allele_dict['_id'])
+            if 'id' in allele_dict:
+                return(allele_dict['id'])
+            elif '_id' in allele_dict:
+                return(allele_dict['_id'])
     return '-'
 
 

pipeline/vr/appendvrids.sh

@@ -14,14 +14,7 @@ SEQ_REPO_DIR="${4:-/usr/local/share/seqrepo}"
 # ----------------------------------------------------
 
 [ `docker ps -f name="seqrepo-rest-service" | wc -l` -gt 1 ] \
-  || docker run \
-       --name seqrepo-rest-service \
-       --user=`id -u`:`id -g` \
-       --detach --rm -p 5000:5000 \
-       --network=host \
-       -v ${SEQ_REPO_DIR}:/usr/local/share/seqrepo \
-       biocommons/seqrepo-rest-service \
-       seqrepo-rest-service /usr/local/share/seqrepo
+    || ../utilities/launch_seqrepo.sh ${SEQ_REPO_DIR}
 
 
 # ...and wait for it to be available
@@ -31,5 +24,6 @@ SEQ_REPO_DIR="${4:-/usr/local/share/seqrepo}"
 # ----------------------------------------------------
 # --- 2. execute append-vr-ids
 # ----------------------------------------------------
-python3 appendVRIds.py -i ${ARTIFACT_DIR}/${INPUT_FILE} \
-       -o ${ARTIFACT_DIR}/${OUTPUT_FILE}
+PATH=../utilities:${PATH} python3 appendVRIds.py \
+    -i ${ARTIFACT_DIR}/${INPUT_FILE} \
+    -o ${ARTIFACT_DIR}/${OUTPUT_FILE}

pipeline/workflow/CompileVCFFiles.py

@@ -762,8 +762,7 @@ def run(self):
             artifacts_dir_host,
             'built_with_priors_clean.tsv',
             'built_with_vr_ids.tsv',
-            self.cfg.vr_docker_image_name,
-            self.cfg.seq_repo_dir
+             self.cfg.seq_repo_dir
         ]
 
         pipeline_utils.run_process(args)